Microsoft Word - numero_41_art_18.docx S. E. Ferreira et alii, Frattura ed Integrità Strutturale, 41 (2017) 129-138; DOI: 10.3221/IGF-ESIS.41.18 129 Focused on Crack Tip Fields A model to quantify fatigue crack growth by cyclic damage accumulation calculated by strip-yield procedures Samuel Elias Ferreira, Jaime Tupiassú Pinho de Castro, Marco Antonio Meggiolaro Pontifical Catholic University of Rio de Janeiro, PUC-Rio, R. Marquês de São Vicente 225, Rio de Janeiro, 22451-900, Brazil ferreirase@hotmail.com, jtcastro@puc-rio.br, meggi@puc-rio.br ABSTRACT. Elber's hypothesis that Keff can be assumed as the driving force for fatigue crack growth (FCG) is the basis for strip-yield models widely used to predict fatigue lives under variable amplitude loads, although it does not explain all load sequence effects observed in practice. To verify if these models are indeed intrinsically better, the mechanics of a typical strip-yield model is used to predict FCG rates based both on Elber's ideas and on the alternative view that FCG is instead due to damage accumulation induced by the cyclic strain history ahead of the crack tip, which does not need or use Keff ideas. The main purpose here is to predict FCG using the cyclic strains induced by the plastic displacements calculated by strip-yield procedures, assuming there are strain limits associated both the with the FCG threshold and with the material toughness. Despite based on conflicting principles, both models can reproduce quite well FCG data, a somewhat surprising result that deserves to be carefully analyzed. KEYWORDS. Fatigue crack growth models; Strip-yield mechanics; Crack closure; Damage accumulation ahead of the crack tip. Citation: Ferreira, S.E., Castro, J.T.P., Meggiolaro, M.A., A model to quantify fatigue crack growth by cyclic damage accumulation calculated by strip-yield procedures, Frattura ed Integrità Strutturale, 41 (2017) 129-138. Received: 28.02.2017 Accepted: 15.04.2017 Published: 01.07.2017 Copyright: © 2017 This is an open access article under the terms of the CC-BY 4.0, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. INTRODUCTION aris and Erdogan clearly demonstrated that stable fatigue crack growth (FCG) rates da/dN can be correlated to stress intensity factor (SIF) ranges K, at least in the central region of typical da/dNK curves, where theirs da/dN  AKm rule applies [1]. Following their idea, many others proposed similar rules to consider the effects of other parameters that can affect FCG rates as well, such as the peak load Kmax or the load ratio R Kmin/Kmax, FCG thresholds Kth(R), and the toughness KC [2]. In particular, after discovering crack closure under tension loads, Elber postulated that additional fatigue damage could only be induced after the crack tip is fully opened under loads greater than Kop, the crack opening load [3-4]. His da/dN  f(Keff  Kmax  Kop if Kop > Kmin) hypothesis can plausibly explain many characteristics of the fatigue cracking behavior, such as FCG delays and arrests induced by overloads (OL), reductions on OL-induced delays after underloads (UL), or the existence of R-dependent FCG thresholds, which can very much affect P S. E. Ferreira et alii, Frattura ed Integrità Strutturale, 41 (2017) 129-138; DOI: 10.3221/IGF-ESIS.41.18 130 fatigue lives under variable amplitude loads (VAL). The Keff idea has been used since then in many semi-empirical FCG models, among them the strip-yield models (SYM) that estimate Kop and FCG lives using a suitable da/dNKeff rule properly fitted to experimental data [5-9]. Works that support the da/dN  f(Keff) hypothesis are extensively reviewed e.g. by Kemp [10] and by Skorupa [11-12], but many other works question it. FCG delays or arrests after OLs under high R while the crack remains fully opened, always maintaining Kmin > Kop [13]; constant FCG rates induced by fixed {K, R}, but highly variable Keff loadings [14-15]; cracks arrested at a given R that restart to grow at a lower R under the same Keff [16]; or the R-insensibility of FCG in inert environments [17], are examples of FCG behaviors that cannot be explained by Elber's postulate. Although this work does not aim to support or to refute Elber's idea, neither to review the works that support or question it, it can be claimed that there is no doubt it still remains controversial. In view of such doubts, this work first uses well-proved strip-yield mechanics [5-9] to model some da/dNK curves measured at low and high R. However, instead of just assuming that a reasonable description of some fatigue data confirms that the Keff hypothesis is valid, the very same mechanics is then used to verify if the same data can be equally described by the alternative view that FCG, instead of controlled by Keff, is due to damage accumulation ahead of the crack tip. This critical damage model (CDM) assumes that fatigue cracks grow by sequentially breaking small volume elements (VE) adjacent to the crack tip after they reach the critical damage the material can sustain. If properly applied, this alternative idea do not need the Keff hypothesis or requires arbitrary data-fitting parameters [13, 18-20]. FCG MODELS our FCG models are studied following: (i) the critical damage model (CDM) proposed in [18-20]; (ii) a strip-yield model (SYM) based on [6]; (iii) a combination of the strip-yield with the critical damage model (SYM-CDM) using fracture mechanics tools; and (iv) a modified strip-yield critical damage model (mod SYM-CDM) proposed here. Fig 1 illustrates the CDM principles that allow the use of N concepts, used to describe fatigue crack initiation, to model FCG as well. This simple model basically assumes that: (i) fatigue cracks grow by successively breaking small VE located ahead of their tips; (ii) such VE can be treated as tiny N specimens fixed along the crack path; (iii) these VE accumulate fatigue damage induced by variable strain ranges, which increase as the crack tip approaches them; and (iv) the fracture of the VE adjacent to the crack tip occurs because it accumulated the entire damage the material can tolerate. Figure 1: Schematics of the FCG process caused by successive fractures of the VE adjacent to the crack tip at every load cycle [2]. F S. E. Ferreira et alii, Frattura ed Integrità Strutturale, 41 (2017) 129-138; DOI: 10.3221/IGF-ESIS.41.18 131 Since constant amplitude SIF range induce constant FCG rates, the VE widths in such cases can also be assumed fixed and equal to the crack increment per cycle. Any given VE suffers damage in each load cycle, caused by the strain loop range induced by that cycle, which depends on the distance xi between the i-th VE and the fatigue crack tip (Fig. 1). The strain ranges acting in any given VE increase at every load cycle, as the crack tip approaches it. The fracture of the VE adjacent to the crack tip occurs when its accumulated damage reaches a critical value, estimated by the linear damage accumulation rule (or by any other suitable damage accumulation rule) as: (1) where Ni(xi) is the number of cycles that the ith VE located at a distance xi from the crack tip would last if only that cycle loading would act during its whole life, and ni is the number of cycles that acted at that load, in this case just one. The CDM uses the elastoplastic strain distribution ahead of the crack tip to calculate damage in every VE. However, like in the LE case, EP models for the stress and strain fields inside the plastic zones pz ahead of a crack tip assumed to have a zero tip radius  = 0, like the HRR field [21-22], are singular at x = 0 as well. To eliminate this undesired and physically inadmissible feature (since no loaded cracks can sustain infinite strains at their tips), the necessarily finite EP stress and strain fields can be estimated by shifting the HRR field origin into the crack by a distance X. This procedure is inspired by Creager and Paris’ idea originally used to estimate stress concentration factors Kt from the SIF of geometrically similar cracks [23]. Under constant SIF range conditions the crack advances a distance equal to the VE width in each cycle (ni  1), so the sum in Eq. (1) can be approximated by an integral along, say, the reverse or cyclic plastic zone (pzr), neglecting in a first approximation fatigue damage induced outside it: (2) The HRR field origin shift can be estimated e.g. assuming X  /2, as Creager and Paris did, where  is the (finite) crack tip radius under the peak load Kmax. To calculate the cyclic plastic strain range p ahead of the crack tip, the modification proposed by Schwalbe [24] can be used as in [19]: (3) where SYc is the cyclic yield strength of the material, E is its Young's modulus, and hc is its Ramberg-Osgood strain- hardening exponent. Since the elastic strain amplitude inside the cyclic plastic zone is neglected in Eq. (3), its associated fatigue life N(x + X) can be estimated from the plastic part of Coffin-Manson’s equation by N,x+X.=,1-2.,,,∆,ε-p.(x+X)-2,ε-c...-,1-c.. (4) where c and c are Coffin-Manson's plastic exponent and coefficient, respectively. Hence, estimating the X displacement of the modified HRR field in the same way as Creager and Paris did with the LE Williams field, i.e. assuming it as   CTOD/2, where CTOD is Schwalbe’s estimate for the Crack Tip Opening Displacement induced by Kmax, then (5) The FCG rate induced by constant SIF ranges, i.e. by fixed {K, R} conditions, can then be estimated by Eq. (2)-(5). Finally, these resulting da/dN values can be used to calculate the constant C in a modified McEvily’s rule that simulates all 3 phases of typical da/dNK curves, considering the FCG threshold Kth and the toughness Kc limits in FCG rates: (6)  ( ) 1i i ii n N x   0 pzrda dx dN N x X          1 12 hcp Yc rx X S E pz x X             2 1 2 1 2 4 2 1 max Yc c KCTOD X E S h               2th c c maxda dN C K K K K K         S. E. Ferreira et alii, Frattura ed Integrità Strutturale, 41 (2017) 129-138; DOI: 10.3221/IGF-ESIS.41.18 132 Another less arbitrary and probably more reasonable way to estimate the X displacement of the HRR field origin (needed to remove the strain singularity at the crack tip) uses the strains induced by Kmax at the crack tip predicted by a suitable strain concentration rule, as described in [19]. In any way, the whole da/dNK curve can be estimated using only well- defined materials properties, without the need for any data-fitting parameter. Such equations describe the simplest formulation of this CDM, since they apply only to constant SIF range conditions, but this model can be further developed to describe FCG under VAL as well, see [20]. Strip-yield models, on the other hand, numerically estimate the Kop needed to find Keff using the classic Dugdale- Barenblatt's idea [25-26], modified to leave a wake of plastically deformed material around the faces of the advancing fatigue crack. Dugdale's model estimates the plastic zone size in a Griffith plate of an elastic perfectly plastic material under plane stress (pl-) conditions, assuming the pz formed ahead of both crack tips under a given Kmax work under a fixed tensile stress equal to the material yield strength SY (neglecting strain-hardening and stress gradients inside the pz). There are several algorithms based on these ideas [5-9]. The SYM algorithm implemented in this work is based in Newman’s original work [6], but it uses the FCG rule and material parameters specified as in the NASGRO code [27]. The validation of this home-made algorithm was performed by comparing its opening stress predictions under several load conditions with results from the literature [28-29]. Newman’s original SYM calculates pz sizes and surface displacements by superposition of two LE problems: (i) a cracked plate loaded by a remote uniform nominal tensile stress and, (ii) a uniform stresses distributed over surface segments near the crack tip. Fig 2 schematizes the crack surface displacements and the stress distributions around the crack tip at the maximum and minimum loads max and min. The plastic zones and the crack wakes left by previous cycles are discretized in a series of rigid-perfectly-plastic 1D bar elements, which are assumed to yield at the flow strength of the material, SFL = (SY + SU)/2, to somehow account for the otherwise neglected strain-hardening effects – a first order approximation. These elements are either intact at the plastic zone or broken at the crack wake, keeping residual plastic deformations. If they are in contact, the broken bar elements can carry compressive stresses, and they can yield in compression when their stresses reach SFL. The elements along the crack face that are not in contact do not affect the crack surface displacements, neither carry stresses. Figure 2: Crack surface displacements and stress distribution along the crack line according to the SYM [6]. It is important to somehow consider the effects of the actually 3D stresses around the crack tip, caused by transversal plastic restrictions induced by the high strain gradients that act there when the plate is thick and cannot be assumed to work under limiting pl- conditions. To do so in the 1D SYM, it uses a thickness-dependent constraint factor  to S. E. Ferreira et alii, Frattura ed Integrità Strutturale, 41 (2017) 129-138; DOI: 10.3221/IGF-ESIS.41.18 133 increase the tensile flow stress SFL in the unbroken elements along the plastic zone during the loading. Hence, this constraint factor should vary from   1 for plane stress conditions in thin plates, to up to   1/(1 2)  3 for plane strain limit conditions in thick plates, where  is Poisson's coefficient (but in practice it is often used as an additional data- fitting parameter). Since there is no crack-tip singularity when the crack closes, this constraint factor is not used to modify the compressive yield strength during unloading, assuming the conditions around the crack tip tend to remain uniaxial. Eq. (7) presented below governs the SYM behavior by requiring compatibility between the LE part of the cracked plate and all bar elements along the crack surfaces and inside the pz ahead of the crack tip. When the length of the wake elements Lj is larger than their displacement Vj under min, they contact and induce a stress j needed to force Vj = Lj. The influence functions f(xi) and g(xi, xj) used in Eq. (7) are related to the plate geometry and its width correction, and are calculated like described in [6]. (7) To improve the resolution of the SYM used here, the pz ahead of the crack tip is divided into 20 bar elements, or twice the number of elements used in the original Newman’s model [6]. After calculating the pz size induced by the peak stress max applied in the current cycle, the widths of the bar elements inside the plastic zone are calculated by Eq. (7), replacing the terms n by 20 and j by SF, see Fig. 2a. When the plate is unloaded to the minimum load min (Fig. 2b), the bar elements inside the pz are also unloaded until some of them near the crack tip start to yield in compression, because they try to reach a stress j ≤ SF. The size of this reverse plastic zone pzr depends on the amount of crack closure and on the transversal constraints induced by the plate thickness. The broken elements located inside the plastic wake formed along the crack surfaces, which store residual deformations, may come into contact and carry compressive stresses as well. Some of these elements may also yield in compression, if they try to reach j ≤ SF. The stresses j at each of the n elements inside the plastic zone and along the plastic wake that surrounds the crack surfaces are calculated solving the system of equations from Eq. 7 using Vi  Li (at max) and n min. Crack opening loads and residual plastic deformations are calculated considering contact stresses. During the crack propagation stage, the opening stress is kept constant during a small arbitrary crack increment to save computer cost. The number of load cycles N required to grow the crack by this increment is calculated by by Eq. (8) [27], in which the parameters Cn, m, p and q are data-fitting constants, Kc is the material toughness, and the FCG threshold Kth can be estimated using a procedure described in [27]. (8) The combination of CDM with SYM procedures uses the same strip-yield description adopted by the SYM to calculate the plastic strain ranges and the consequent fatigue damage distribution ahead of the crack tip. This replace the displaced HRR strain field used by the original critical damage model [19]. This model estimates the FCG increments in a cycle-by- cycle basis by a gradual damage accumulation process, but considering possible crack closure effects on the cyclic strain field ahead of the crack tip. The pz ahead of the crack tip is divided into small bar elements with constant width, whose quantity varies between 150 and 550, depending on the loading conditions. The plastic displacements within the SYM strip-yield are transformed into plastic strains using a solution proposed by Rice [30] to estimate the strain field based on CTOD variations, properly modified to consider each bar element displacements by (9) The displacements Lmax and Lmin of the ith element inside the pz are calculated at the maximum and minimum applied stresses. The position of the elements starting from the crack tip, xct(i), is located at the center of each element. The strain range y that acts at each element center can be correlated to the number of cycles that would be required to break that element by the plastic part of Coffin-Manson’s rule (C&M) Eq. (10), or else by SWT rule Eq. (11), to consider peak stress effects. Notice that only the plastic part of the strain range can be considered by this SY-CDM, because the strain ranges estimated from the SYM displacements are calculated assuming rigid-perfectly-plastic bar elements. The total damage accumulated by each element is evaluated by Palmgren-Miner’s rule Eq. (1).     1 , n i n i j i j j V f x g x x            1 1m p qn eff th max cda dN C K K K K K         ( ) ( ) ( )2 2 ( ) ( )y max ct min ctlog L x Li i i i ix      S. E. Ferreira et alii, Frattura ed Integrità Strutturale, 41 (2017) 129-138; DOI: 10.3221/IGF-ESIS.41.18 134 (10) (11) Since the numerical model calculates fatigue damage at the central position of each VE, the crack increment induced by each load cycle is given by the location where the damage reaches one ahead of the crack tip. Therefore, the damage value of the new first element are calculated using a linear interpolation procedure. As the number of elements is unchanged, to keep the width sum of all elements equal to the pz size, the first and the last element can have a variable width. Moreover, since Eq. (6) reproduces the sigmoidal shape of da/dNK curves, and since its only adjustable constant C can be directly calculated for any {K, R} combination from the N properties of the material, the CDM used here in fact have no data- fitting parameters (whereas the NASGRO FCG rule used at the SYM needs 4 of them). The C constant (Eq. 6) is found using several da/dN values calculated by SY-CDM procedures using C&M or SWT rules. This process is similar to the adopted by the original CDM. The main difference here is the replacement of a shifted HRR strain field by a strain field derived from the displacement field of the SYM. The new modification proposed here for the SY-CDM eliminates the need of assuming that the da/dN versus K curve of the material is always described by McEvily’s rule Eq. (6). This can be achieved by assuming two new hypotheses, which are also based on the physics of the FCG process. The first supposes that there is a limit strain range below which the crack does not grow by fatigue, which is directly related to the SIF range threshold Kth. The second assumes the crack becomes unstable at a maximum plastic strain related to the critical stress intensity factor, or to the material toughness. These hypotheses are described by: (12) The plastic strain range calculated by Eq. (12) is used by the modified SY-CDM to calculate the damage at each element by Eq. (10) or (11). The interpolation routine was improved to work with a fixed number of elements (400) for any load condition and, due these hypotheses, one less step is required to estimate the FCG Finally, a major fringe-benefit of all CDMs used in this work must be emphasized: if they can reasonably estimate FCG rates for a given material, they do so using only its N, Kth, and KC properties, without the need for any adjustable data-fitting parameters. Therefore, these simple and sound models can indeed be called predictive, since they do not need or use actual FCG data points to estimate da/dN rates. The results presented next support this claim. EXPERIMENTAL RESULTS AND DISCUSSIONS hese four models are compared against experimental da/dNK data measured at R  0.1 and R  0.7 for two materials, a 7075-T6 Al alloy and a 1020 low carbon steel, as described elsewhere [19]. These materials properties and the C values used by the C&P CDM (model i) are listed in Tab. 1. FCG rates predictions by the SYM (model ii) use material properties from NASGRO version 4.02, as shown in Tab. 2. But instead of using Poisson's coefficients to estimate the constraint factor , its value was varied to verify its effect on FCG predictions. Values  3 for 7075 and  2 for the 1020 resulted in good approximations and are adopted here. The Kth(R) and KIC from Tab. 1 are used in Eq. 8, since they approximate the data better than the values listed in NASGRO. The constants C from Eq. (6), model (iii), are listed in Tab. 3. Material SY (MPa) SU (MPa) c (MPa) c b c KIC (MPam) Kth (MPam) C (for K in MPam) R  0.1 R  0.7 7075-T6 498 576 709 0.12 -0.056 -0.75 25.4 3.4 2.9 8.23e-09 1020 285 491 815 0.25 -0.114 -0.54 277 11.6 7.5 2.73e-10 Table 1: Material properties [19] and C values obtained by several strain concentration rules.   1( )) (1 2 2 cy cN ii        1 max( ) 1 2 ( ) ( ) 2 b c y c cN i i i        ,mod , , , ,max( ) ( ) ( ) ( ) ( ) ( )y y y th y cr y cr yi i i i i i                 T S. E. Ferreira et alii, Frattura ed Integrità Strutturale, 41 (2017) 129-138; DOI: 10.3221/IGF-ESIS.41.18 135 Material SY (MPa) SU (MPa) Cn (for K in MPamm m p q 7075-T6 (M7HA03AB1) 461.9 524 9.68610-12 3 0.5 1 1015-1026 (C1BB11AB1) 262 399.9 1.51510-14 3.7 0.5 0.5 Table 2: Properties and Forman-Newman parameters from the NASGRO 4.02 database [27]. N equation C (for K in MPam) 7075-T6 1020 C&M 8.73e-09 2.64e-09 SWT 1.49e-08 4.50e-09 Table 3: Constant C for modified CDMs obtained from SYM-calculated cyclic strain fields. Figs. 3 and 4 show the measured da/dNK points and the curves predicted by the original CDM based on Creager and Paris (C&P), by the original SYM (assuming   3), by the SY-CDMs (SY-C&M and SY-SWT), and also by the modified SY-CDMs proposed here (SY-C&M modified and SY-SWT modified). The FCG data has been obtained under R  0.1 and R  0.7 using standard ASTM E647 procedures for a 7075-T6 Al alloy and a 1020 steel, respectively. Recall that the SY-CDM curves are predicted from the N damage induced by the cyclic strain fields generated by SYM strip-yield procedures using Coffin-Manson or STW N rules. These models use only the plastic part of those N rules because the SYM numerical procedures discretize the pz ahead of the crack tip using rigid-perfectly-plastic VE elements. Recall as well that the modified SY-CDM proposed here does not need to use a previously chosen da/dNK rule (Eq. 6) due the two limiting strain values introduced in the model, one for the FCG threshold and the other for the toughness. Figure 3: Strip-yield and critical damage models for the 7075-T6 at R = 0.1 and 0.7. S. E. Ferreira et alii, Frattura ed Integrità Strutturale, 41 (2017) 129-138; DOI: 10.3221/IGF-ESIS.41.18 136 Notice in Fig. 3 that the da/dNK curve estimated by the SY-C&M model is essentially equal to the curve generated by the original C&P CDM model. Both estimates are quite reasonable for R 0.1, albeit not as good for R  0.7. The SYM curve (generated assuming   3) describes better the data points measured at R  0.7. The SWT N model estimates higher FCG rates than the C&M model for both R-ratios, as expected. The model proposed here (SY-C&M-modified, which does not need an assumed FCG rule) yielded the best estimates at R  0.1 and had a performance similar to the SYM at R  0.7. The modified SY-CDMs had in particular a better performance at the higher K ranges, where the original models systematically estimated FCG rates higher than the data. The original CDM [19] and SY-CDM models need a pre-chosen McEvily’-type da/dNK curve, whose single adjustable parameter can however be calculated by N procedures. Their good performance certainly is not a coincidence, since they use no adjustable data-fitting parameters and their predictions are entirely based on measured N properties. In fact, when compared to SYM estimates based on Keff concepts and on a FCG rule that has 4 adjustable data-fitting parameters, not to mention the constraint factor  that in practice is frequently used as a 5th adjustable parameter, the CDM performance could be even qualified as quite impressive for a so simple model. The results for the 1020 steel are shown in Fig. 4. The original CDM based on a Creager and Paris shift of the HRR field reproduced well the data trend, but yielded slight non-conservative FCG estimates at R  0.1. For R  0.7 it presented a still better performance. The original SYM had a similar performance at R  0.1, but instead generated slight conservative predictions, which deviated from data at low K values. For R  0.7 its predictions were not good. The modified SY- C&M generated quite reasonable predictions for R  0.7, but for R  0.1 they were maybe too conservative. The other models yielded too conservative predictions for both R-ratios. Figure 4: Strip-yield and critical damage models for the 1020 steel at R = 0.1 and 0.7. Two facts resulting from this academic exercise must be emphasized. First, their FCG estimates were quite reasonable, an indication that their ideas about the mechanics of the FCG process probably are reasonable as well. This at least may be seen as an indication that the procedures used in these simple models are at least coherent, a reassuring evidence. However, the second fact is still more interesting, since it could not be anticipated. The study presented above show that S. E. Ferreira et alii, Frattura ed Integrità Strutturale, 41 (2017) 129-138; DOI: 10.3221/IGF-ESIS.41.18 137 FCG rates estimated by opposing ideas can yield similarly reasonable results. Moreover, when the SYM and the CDM techniques are properly combined, they also generate reasonable predictions. This does not means that these methods are equivalent. Indeed, while the CDM FCG rate estimates requires only measurable N properties and need no data-fitting parameters, the SYM estimates use at least four data-fitting parameters in a pre-chosen FCG rule to achieve similar results. This is probably not a major problem from an engineering point of view, if such parameters are available. However, it is undeniable that to generate similar descriptions of measured FCG data using only standard N properties and basic mechanical principles is a point in favor of CDM ideas. Finally, an additional point must be emphasized as well, although it is more philosophical than operational: the results presented here also indicate that a good description of some experimental data cannot be claimed as a conclusive proof of any model suitability, let alone of its prevalence. What is really important when discussing the performance of any given model is to clearly identify which set of properly measured experimental data it cannot describe well. This point is important, since after so many years still there is no consensus even about which are the true fatigue crack driving forces, let alone on the best FCG model. Indeed, whereas many defend that fatigue cracks are driven by Keff, others affirm that fatigue crack closure is not even a major issue in FCG. The authors hope this relatively straightforward modeling exercise can contribute at least to avoid the radical opinions that are still too common in this field. CONCLUSIONS CG models based on critical damage and strip-yield procedures are used to estimate da/dNK curves of two materials under low and high R-ratios. These models are based on contradictory hypotheses about the cause for the FCG behavior. Whereas the SYMs assume FCG is driven by Keff, so that it depends on the interference of the plastic wakes left behind the crack tip along the crack surfaces, the CDMs suppose fatigue cracks propagate by sequentially breaking volume elements ahead of the crack tip, which fail because they accumulate all the fatigue damage they could sustain. All FCG models studied here are compared against properly measured da/dNK curves of a 7075-T6 Al alloy and of an AISI 1020 low carbon steel, which were experimentally obtained following standard ASTM E647 procedures. Moreover, the N properties of such materials were also measured by standard procedures, following ASTM E606 recommendations. Moreover, both the FCG and the fatigue crack initiation properties were measured in coupons machined from the same material lot, to avoid any inconsistency in the data. Both the original CDMs (based on the HRR field displaced to eliminate the strain singularity ate the crack tip) and SYMs can describe reasonably well the measured data, even though they are apparently contradictory from a conceptual point of view. This is certainly an indication that such models are not incompatible. This claim is verified here by mixing them, using the strip yield mechanics instead of the HRR field to generate the strain field ahead of the crack tip needed for the critical damage calculations. First, this is done maintaining the hypothesis that the FCG curves can be described by McEvily’s single parameter model. Then, two new hypotheses are proposed to eliminate the need for such an assumption, namely (i) there is a limit strain range related to the threshold stress intensity factor range, and (ii) there is a maximum plastic strain related to the critical stress intensity factor. These limit strain values can be introduced in the N damage calculations, eliminating one calculation step and allowing the CDM to easier deal with variable amplitude loading problems, a feature that will be discussed in future works. Finally, the quite reasonable performance of the predictions obtained from models based on so different hypothesis about the FCG driving forces also indicates that the good fitting of some properly obtained data set is not enough to prove which one is the best. REFERENCES [1] Paris, P.C., Erdogan, F. A critical analysis of crack propagation laws. J Basic Eng 85 (1963) 528-534. [2] Castro, J.T.P., Meggiolaro, M.A., Fatigue Design Techniques, volume 3: Crack Propagation, Temperature and Statistical Effects. CreateSpace (2016). [3] Elber, W., Fatigue crack closure under cyclic tension. Eng Fract Mech, 2 1970) 37-45. [4] Elber, W., The significance of fatigue crack closure. Damage Tolerance in Aircraft Structures, ASTM STP, 486 (1971) 230-242. F S. E. Ferreira et alii, Frattura ed Integrità Strutturale, 41 (2017) 129-138; DOI: 10.3221/IGF-ESIS.41.18 138 [5] Dill, H.D., Saff, C.R., Spectrum crack growth prediction method based on crack surface displacement and contact analyses. Fatigue Crack Growth under Spectrum Loads, ASTM STP 595 (1976) 306-319. [6] Newman, Jr, JC., A crack-closure model for predicting fatigue crack growth under aircraft spectrum loading. Methods and Models for Predicting Fatigue Crack Growth under Random Loading, ASTM STP 748 (1981) 53-84. [7] de Koning, A.U., Liefting, G.. Analysis of crack opening behavior by application of a discretized strip yield model. Mechanics of Fatigue Crack Closure, ASTM STP 982 (1988) 437-458. [8] Wang, G.S., Blom, A.F., A strip model for fatigue crack growth predictions under general load conditions. Eng Fract Mech, 40 (1991) 507-533. [9] Beretta,S, Carboni,M. A strip-yield algorithm for the analysis of closure evaluation near the crack tip. Eng Fract Mech 72 (2005) 1222-1237. [10] Kemp, P.M.J., Fatigue crack closure – a review. TR90046, Royal Aerospace Establishment, UK, (1990). [11] Skorupa, M., Load interaction effects during fatigue crack growth under variable amplitude loading - a literature review - part I: empirical trends, Fatigue Fract Eng Mater Struct, 21 (1998) 987-1006. [12] Skorupa, M., Load interaction effects during fatigue crack Growth under variable amplitude loading - a literature review - part II: qualitative interpretation, Fatigue Fract Eng Mater Struct, 22 (1999) 905-926. [13] Castro, J.T.P., Meggiolaro, M.A., Miranda, A.C.O., Singular and non-singular approaches for predicting fatigue crack growth behavior. Int J Fatigue, 27 (2005) 1366-1388. [14] Castro, J.T.P., Meggiolaro, M.A., González, J.A.O., Can Keff be assumed as the driving force for fatigue crack growth? Frattura ed Integrità Strutturale, 33 (2015) 97-104. [15] Castro, J.T.P., González, J.A.O., Meggiolaro, M.A., González, G.L.G., Freire, J.L.F., Some questions about assuming Keff as the sole FCG driving force. Submitted to Int J Fatigue, (2016). [16] Chen, D.L., Weiss, B., Stickler, R., The effective fatigue threshold: significance of the loading cycle below the crack opening load. Int J Fatigue, 16 (1994) 485-491. [17] Vasudevan, A. K., Sadananda, K., Holtz, R. L., Analysis of vacuum fatigue crack growth results and its implications, International J. Fatigue, 27 (2005) 1519-1529. [18] Castro, J.T.P., Kenedi, P.P., Prediction of fatigue crack growth rates departing from Coffin-Manson concepts. J Braz Soc Mech Sci Eng, 17 (1995) 292-303 (in Portuguese). [19] Durán, J.A.R., Castro, J.T.P., Payão Filho, J.C., Fatigue crack propagation prediction by cyclic plasticity damage accumulation models, Fatigue Fract Eng Mater Struct, 26 (2003) 137-150. [20] Castro, J.T.P., Meggiolaro, M.A., Miranda, A.C.O., Fatigue crack growth predictions based on damage accumulation calculations ahead of the crack tip, Comput Mat Sci, 46 (2009) 115-123. [21] Rice, J.R., Rosengren, G.F., Plane strain deformation near a crack tip in a power-law hardening material, J Mech Phys Solids, 16 (1968) 1-12. [22] Hutchinson, J.W., Singular behavior at the end of a tensile crack tip in a hardening material, J Mech Phys Solids 16 (1968) 13-31. [23] Creager, M., Paris, P.C., Elastic field equations for blunt cracks with reference to stress corrosion cracking, Int J Fract Mech, 3 (1967) 247-252. [24] Schwalbe, K.H., Comparison of several fatigue crack propagation laws with experimental results, Eng Fract Mech 6 (1974) 325-341. [25] Dugdale, D. S., Yielding of sheets containing slits, J Mech Phys Solids, 8 (1960) 100-104. [26] Barenblatt, G.I., The mathematical theory of equilibrium cracks in brittle fracture. Advances in Applied Mechanics, 7 (1962) 55-192. [27] NASGRO – Fracture Mechanics and Fatigue Crack Growth Analysis Software, Reference Manual, version 4.02, (2002). [28] Newman, J.C., FASTRAN II: a fatigue crack growth structural analysis program, NASA Technical Memorandum 104159, LRC Hampton, (1992). [29] Newman, J.C., A Crack Opening Stress Equation for Fatigue Crack Growth. Int J Fract, 24 (1984) R131-R135. [30] Rice, J.R., Mechanics of crack tip deformation and extension by fatigue. Fatigue Crack Propagation, ASTM STP 415 (1967) 247-311. << /ASCII85EncodePages false /AllowTransparency false /AutoPositionEPSFiles true /AutoRotatePages /None /Binding /Left /CalGrayProfile (Dot Gain 20%) /CalRGBProfile (sRGB IEC61966-2.1) /CalCMYKProfile (U.S. Web Coated \050SWOP\051 v2) /sRGBProfile (sRGB IEC61966-2.1) /CannotEmbedFontPolicy /Error /CompatibilityLevel 1.4 /CompressObjects /Tags /CompressPages true /ConvertImagesToIndexed true /PassThroughJPEGImages true /CreateJobTicket false /DefaultRenderingIntent /Default /DetectBlends true /DetectCurves 0.0000 /ColorConversionStrategy /CMYK /DoThumbnails false /EmbedAllFonts true /EmbedOpenType false /ParseICCProfilesInComments true /EmbedJobOptions true /DSCReportingLevel 0 /EmitDSCWarnings false /EndPage -1 /ImageMemory 1048576 /LockDistillerParams false /MaxSubsetPct 100 /Optimize true /OPM 1 /ParseDSCComments true /ParseDSCCommentsForDocInfo true /PreserveCopyPage true /PreserveDICMYKValues true /PreserveEPSInfo true /PreserveFlatness true /PreserveHalftoneInfo false /PreserveOPIComments true /PreserveOverprintSettings true /StartPage 1 /SubsetFonts true /TransferFunctionInfo /Apply /UCRandBGInfo /Preserve /UsePrologue false /ColorSettingsFile () /AlwaysEmbed [ true ] /NeverEmbed [ true ] /AntiAliasColorImages false /CropColorImages true /ColorImageMinResolution 300 /ColorImageMinResolutionPolicy /OK /DownsampleColorImages true /ColorImageDownsampleType /Bicubic /ColorImageResolution 300 /ColorImageDepth -1 /ColorImageMinDownsampleDepth 1 /ColorImageDownsampleThreshold 1.50000 /EncodeColorImages true /ColorImageFilter /DCTEncode /AutoFilterColorImages true /ColorImageAutoFilterStrategy /JPEG /ColorACSImageDict << /QFactor 0.15 /HSamples [1 1 1 1] /VSamples [1 1 1 1] >> /ColorImageDict << /QFactor 0.15 /HSamples [1 1 1 1] /VSamples [1 1 1 1] >> /JPEG2000ColorACSImageDict << /TileWidth 256 /TileHeight 256 /Quality 30 >> /JPEG2000ColorImageDict << /TileWidth 256 /TileHeight 256 /Quality 30 >> /AntiAliasGrayImages false /CropGrayImages true /GrayImageMinResolution 300 /GrayImageMinResolutionPolicy /OK /DownsampleGrayImages true /GrayImageDownsampleType /Bicubic /GrayImageResolution 300 /GrayImageDepth -1 /GrayImageMinDownsampleDepth 2 /GrayImageDownsampleThreshold 1.50000 /EncodeGrayImages true /GrayImageFilter /DCTEncode /AutoFilterGrayImages true /GrayImageAutoFilterStrategy /JPEG /GrayACSImageDict << /QFactor 0.15 /HSamples [1 1 1 1] /VSamples [1 1 1 1] >> /GrayImageDict << /QFactor 0.15 /HSamples [1 1 1 1] /VSamples [1 1 1 1] >> /JPEG2000GrayACSImageDict << /TileWidth 256 /TileHeight 256 /Quality 30 >> /JPEG2000GrayImageDict << /TileWidth 256 /TileHeight 256 /Quality 30 >> /AntiAliasMonoImages false /CropMonoImages true /MonoImageMinResolution 1200 /MonoImageMinResolutionPolicy /OK /DownsampleMonoImages true /MonoImageDownsampleType /Bicubic /MonoImageResolution 1200 /MonoImageDepth -1 /MonoImageDownsampleThreshold 1.50000 /EncodeMonoImages true /MonoImageFilter /CCITTFaxEncode /MonoImageDict << /K -1 >> /AllowPSXObjects false /CheckCompliance [ /None ] /PDFX1aCheck false /PDFX3Check false /PDFXCompliantPDFOnly false /PDFXNoTrimBoxError true /PDFXTrimBoxToMediaBoxOffset [ 0.00000 0.00000 0.00000 0.00000 ] /PDFXSetBleedBoxToMediaBox true /PDFXBleedBoxToTrimBoxOffset [ 0.00000 0.00000 0.00000 0.00000 ] /PDFXOutputIntentProfile () /PDFXOutputConditionIdentifier () /PDFXOutputCondition () /PDFXRegistryName () /PDFXTrapped /False /CreateJDFFile false /Description << /ARA /BGR /CHS /CHT /CZE /DAN /DEU /ESP /ETI /FRA /GRE /HEB /HRV (Za stvaranje Adobe PDF dokumenata najpogodnijih za visokokvalitetni ispis prije tiskanja koristite ove postavke. Stvoreni PDF dokumenti mogu se otvoriti Acrobat i Adobe Reader 5.0 i kasnijim verzijama.) /HUN /ITA /JPN /KOR /LTH /LVI /NLD (Gebruik deze instellingen om Adobe PDF-documenten te maken die zijn geoptimaliseerd voor prepress-afdrukken van hoge kwaliteit. De gemaakte PDF-documenten kunnen worden geopend met Acrobat en Adobe Reader 5.0 en hoger.) /NOR /POL /PTB /RUM /RUS /SKY /SLV /SUO /SVE /TUR /UKR /ENU (Use these settings to create Adobe PDF documents best suited for high-quality prepress printing. Created PDF documents can be opened with Acrobat and Adobe Reader 5.0 and later.) >> /Namespace [ (Adobe) (Common) (1.0) ] /OtherNamespaces [ << /AsReaderSpreads false /CropImagesToFrames true /ErrorControl /WarnAndContinue /FlattenerIgnoreSpreadOverrides false /IncludeGuidesGrids false /IncludeNonPrinting false /IncludeSlug false /Namespace [ (Adobe) (InDesign) (4.0) ] /OmitPlacedBitmaps false /OmitPlacedEPS false /OmitPlacedPDF false /SimulateOverprint /Legacy >> << /AddBleedMarks false /AddColorBars false /AddCropMarks false /AddPageInfo false /AddRegMarks false /ConvertColors /ConvertToCMYK /DestinationProfileName () /DestinationProfileSelector /DocumentCMYK /Downsample16BitImages true /FlattenerPreset << /PresetSelector /MediumResolution >> /FormElements false /GenerateStructure false /IncludeBookmarks false /IncludeHyperlinks false /IncludeInteractive false /IncludeLayers false /IncludeProfiles false /MultimediaHandling /UseObjectSettings /Namespace [ (Adobe) (CreativeSuite) (2.0) ] /PDFXOutputIntentProfileSelector /DocumentCMYK /PreserveEditing true /UntaggedCMYKHandling /LeaveUntagged /UntaggedRGBHandling /UseDocumentProfile /UseDocumentBleed false >> ] >> setdistillerparams << /HWResolution [2400 2400] /PageSize [612.000 792.000] >> setpagedevice